· Jul 16, 2026
xAI introduces Grok 4.5 for coding and knowledge work
Markdown · JSON · Original source
’s page, published July 16, 2026, announces 4.5 as a model for coding, tasks and knowledge work. xAI says it was trained alongside and is its “strongest model ever.” The page argues that training focused on technical work, including multi-step software engineering, and says the model excels at real engineering tasks. Its results are mixed: Grok 4.5 leads the listed models on SWE Marathon, while other models score higher on several listed tests. xAI says competitor figures come from developers’ system cards or benchmark leaderboards; these comparisons do not by themselves show how the model performs in varied workplace use. The announcement also emphasizes speed, output-token efficiency and price. xAI reports an average of 15,954 per SWE Bench Pro task, compared with 67,020 for Opus 4.8 (max), and says Grok 4.5 costs $2 per million input tokens and $6 per million output tokens. It presents the combination as “the highest intelligence per unit of and cost.” xAI says the model is available in Grok Build, Cursor and the SpaceXSI console, and that limited-time free usage is offered in Grok Build and Cursor. Whether the benchmark and efficiency results translate into dependable performance on users’ own tasks remains unknown.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
· Article
Source details
Original source date: Jul 16, 2026 (2026-07-16T00:00:00Z).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-06T08:33:24.128Z. First observed: 2026-10-18T07:29:21.396Z. Retrieved: 2026-10-24T00:13:43.876Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 24, 2026; originally published July 16, 2026. says competitor figures are drawn from developers’ published system cards or leaderboards; the page does not establish how benchmark results translate to performance on users’ own tasks. The page presents company claims and benchmark results; it does not provide independent verification of the claimed capabilities or efficiency.
Open question
How does 4.5 perform on users’ own coding and office-work tasks, beyond the lists?
Source-stated
xAI describes Grok 4.5 as intended for coding, agentic tasks and knowledge work, and calls it its strongest model to date.
Today, we're launching Grok 4.5, SpaceXSI's smartest model built to excel at coding, agentic tasks, and knowledge work. It's our strongest model ever
Source-stated
The benchmark results shown are mixed: Grok 4.5 leads the listed models on SWE Marathon, while other listed models score higher on several other tests.
SWE Marathon resolution rate (pass@1): Grok 4.5 29.0%, Opus 4.8 (max) 26.0%, Fable (max) 24.0%
Source-stated
xAI says Grok 4.5 averages fewer output tokens than Opus 4.8 (max) on SWE Bench Pro tasks.
Grok 4.5 resolves tasks with 15,954 output tokens on average, about 4.2× fewer than Opus 4.8 (max) at 67,020
Source-stated
xAI lists prices of $2 per million input tokens and $6 per million output tokens.
Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens.
Source-stated
xAI says Grok 4.5 is available in Grok Build, Cursor and the SpaceXSI console, and offers free use for a limited time in Grok Build and Cursor.
We’re offering free Grok 4.5 usage for a limited time in Grok Build and Cursor.
Source and contributor context
xAI · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Agentic AI
AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Model tokens
Small chunks of text, often parts of words, that a language model reads and writes; usage and prices are often counted in tokens.
xAI
xAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Grok
Grok is a family of AI models from xAI, with capability claims and evaluations published by its developer.
Cursor
Cursor makes an AI coding environment and publishes evaluations of agents working on software tasks.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 221dcd9c-4e2e-6945-bed2-538d3d29d854.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-06T08:28:46.088Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: 12d8f6691919ad2c111485b403fe16cfcd92cbec2abee294d5b7714c6862ef18. Source text hash: a88a8c0548b60f6b7904667d52dcfdbd553e2f9fae7ea14827046e5a4300ca0b.
Revision and source-check history
Version 1 · 2026-11-06T08:33:24.128Z
xAI introduces Grok 4.5 for coding and knowledge work
’s page, published July 16, 2026, announces 4.5 as a model for coding, tasks and knowledge work. xAI says it was trained alongside and is its “strongest model ever.” The page argues that training focused on technical work, including multi-step software engineering, and says the model excels at real engineering tasks. Its results are mixed: Grok 4.5 leads the listed models on SWE Marathon, while other models score higher on several listed tests. xAI says competitor figures come from developers’ system cards or benchmark leaderboards; these comparisons do not by themselves show how the model performs in varied workplace use. The announcement also emphasizes speed, output-token efficiency and price. xAI reports an average of 15,954 per SWE Bench Pro task, compared with 67,020 for Opus 4.8 (max), and says Grok 4.5 costs $2 per million input tokens and $6 per million output tokens. It presents the combination as “the highest intelligence per unit of and cost.” xAI says the model is available in Grok Build, Cursor and the SpaceXSI console, and that limited-time free usage is offered in Grok Build and Cursor. Whether the benchmark and efficiency results translate into dependable performance on users’ own tasks remains unknown.
Original source
· Article
Source details
Original source date: Jul 16, 2026 (2026-07-16T00:00:00Z).
Date basis: Date stated on the publisher’s page.
Added to Escape Meridian: 2026-11-06T08:33:24.128Z. First observed: 2026-10-18T07:29:21.396Z. Retrieved: 2026-10-24T00:13:43.876Z.
Limits
Text as retrieved on October 24, 2026; originally published July 16, 2026. says competitor figures are drawn from developers’ published system cards or leaderboards; the page does not establish how benchmark results translate to performance on users’ own tasks. The page presents company claims and benchmark results; it does not provide independent verification of the claimed capabilities or efficiency.
Open question
How does 4.5 perform on users’ own coding and office-work tasks, beyond the lists?
Source-stated
xAI describes Grok 4.5 as intended for coding, agentic tasks and knowledge work, and calls it its strongest model to date.
Today, we're launching Grok 4.5, SpaceXSI's smartest model built to excel at coding, agentic tasks, and knowledge work. It's our strongest model ever
Source-stated
The benchmark results shown are mixed: Grok 4.5 leads the listed models on SWE Marathon, while other listed models score higher on several other tests.
SWE Marathon resolution rate (pass@1): Grok 4.5 29.0%, Opus 4.8 (max) 26.0%, Fable (max) 24.0%
Source-stated
xAI says Grok 4.5 averages fewer output tokens than Opus 4.8 (max) on SWE Bench Pro tasks.
Grok 4.5 resolves tasks with 15,954 output tokens on average, about 4.2× fewer than Opus 4.8 (max) at 67,020
Source-stated
xAI lists prices of $2 per million input tokens and $6 per million output tokens.
Grok 4.5 is priced at $2 per million input tokens and $6 per million output tokens.
Source-stated
xAI says Grok 4.5 is available in Grok Build, Cursor and the SpaceXSI console, and offers free use for a limited time in Grok Build and Cursor.
We’re offering free Grok 4.5 usage for a limited time in Grok Build and Cursor.
Source and contributor context
xAI · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Agentic AI
AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
Benchmark
A defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Model tokens
Small chunks of text, often parts of words, that a language model reads and writes; usage and prices are often counted in tokens.
xAI
xAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Grok
Grok is a family of AI models from xAI, with capability claims and evaluations published by its developer.
Cursor
Cursor makes an AI coding environment and publishes evaluations of agents working on software tasks.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 221dcd9c-4e2e-6945-bed2-538d3d29d854.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-06T08:28:46.088Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model, second_opinion_categorical.
Observation: 12d8f6691919ad2c111485b403fe16cfcd92cbec2abee294d5b7714c6862ef18. Source text hash: a88a8c0548b60f6b7904667d52dcfdbd553e2f9fae7ea14827046e5a4300ca0b.
passed · 2026-11-06T08:28:46.088Z · anthropic/claude-opus-5.5
Publication: 221dcd9c-4e2e-6945-bed2-538d3d29d854
attribution — pass: The reading attributes the announcement and its claims to xAI's own page at x.ai, which is the publisher. Capability, benchmark, efficiency and price claims are framed as xAI's statements, not as established facts. The page calls the model SpaceXSI's, and the reading keeps that name for the console.
Today, we're launching Grok 4.5, SpaceXSI's smartest model built to excel at coding, agentic tasks, and knowledge work.
summary — pass: The title, summary and question faithfully convey the launch, the training focus, the mixed benchmark picture, token efficiency, pricing and availability. Benchmark, agentic and tokens are defined beside the reading. Names such as SWE Marathon, Cursor and Opus 4.8 are context-clear enough not to block understanding. The question raises a real open issue that follows from the evidence.
Grok 4.5 excels at real engineering tasks and exceeds comparable leading models at these tasks.
date — pass: The reading gives July 16, 2026, which matches the UTC calendar day of evidence.originalPublished. Evidence.kind is publisher-page, so no archive disclosure is needed. The reading uses no hindsight framing and presents no later events. Availability and free usage are reported as xAI's statements at launch.
Grok 4.5 is available today in Grok Build, in Cursor on all plans, and from the SpaceXSI console.
support — pass: Each claim's excerpt is verbatim and supports it. The mixed-results characterization matches the charts, where Grok 4.5 leads only SWE Marathon. The token figures, prices and the quoted phrase about intelligence per unit of time and cost appear in the text. The caveat about real-world performance is framed as unknown, not asserted, and no unsupported benefits are implied.
Overall, Grok 4.5 delivers the highest intelligence per unit of time and cost.
media — pass: No media is selected, and no verified media candidates were supplied, so nothing was misidentified or invented.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
xAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIxAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIGrok is a family of AI models from xAI, with capability claims and evaluations published by its developer.
Explore GrokAI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
Definition referenceCursor makes an AI coding environment and publishes evaluations of agents working on software tasks.
Explore CursorA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceSmall chunks of text, often parts of words, that a language model reads and writes; usage and prices are often counted in tokens.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMExAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIxAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceGrok is a family of AI models from xAI, with capability claims and evaluations published by its developer.
Explore GrokA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referencexAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIxAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIGrok is a family of AI models from xAI, with capability claims and evaluations published by its developer.
Explore GrokAI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.
Definition referenceCursor makes an AI coding environment and publishes evaluations of agents working on software tasks.
Explore CursorA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceSmall chunks of text, often parts of words, that a language model reads and writes; usage and prices are often counted in tokens.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMExAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIxAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAIA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referenceGrok is a family of AI models from xAI, with capability claims and evaluations published by its developer.
Explore GrokA defined set of tests used to compare systems. Results depend on the tasks and measures selected.
Definition referencexAI develops the Grok family of AI models and publishes its own release claims and evaluations.
Explore xAI